Minireview: Mechanisms of Growth Hormone-Mediated Gene Regulation

Abstract

GH exerts a diverse array of physiological actions that include prominent roles in growth and metabolism, with a major contribution via stimulating IGF-1 synthesis. GH achieves its effects by influencing gene expression profiles, and Igf1 is a key transcriptional target of GH signaling in liver and other tissues. This review examines the mechanisms of GH-mediated gene regulation that begin with signal transduction pathways activated downstream of the GH receptor and continue with chromatin events at target genes and additionally encompasses the topics of negative regulation and cross talk with other cellular inputs. The transcription factor, signal transducer and activator of transcription 5b, is regarded as the major signaling pathway by which GH achieves its physiological effects, including in stimulating Igf1 gene transcription in liver. Recent studies exploring the mechanisms of how activated signal transducer and activator of transcription 5b accomplishes this are highlighted, which begin to characterize epigenetic features at regulatory domains of the Igf1 locus. Further research in this field offers promise to better understand the GH-IGF-1 axis in normal physiology and disease and to identify strategies to manipulate the axis to improve human health.

GH exerts a diverse array of physiological actions that includes major roles in growth and metabolism. The original somatomedin hypothesis of Salmon and Daughaday (1) proposed that a serum factor controlled by GH, or somatomedin, was directly responsible for stimulating sulfate incorporation by cartilage, and several lines of investigation have since identified IGF-1 as the principal somatomedin responsible for this effect (2). In support of the somatomedin hypothesis, many of the characteristic phenotypic features of GH deficiency are indeed recapitulated with mutations of the gene encoding IGF-1 in both rodents and humans (3–5), although there is now a much better recognition of the role of local tissue-derived IGF-1 production, rather than strictly endocrine-acting serum IGF-1, on mediating many of the effects of GH (2). GH treatment restores growth velocity in children with GH deficiency (6), whereas replacement therapy in adults improves body composition, exercise capacity, skeletal integrity, and quality of life measures (7). GH has been purported to have an antiaging effect (8), but diminished signaling of the GH-IGF-1 axis has been associated with longevity across many species (9), intimating deleterious effects from this pathway. Indeed this phenomenon extends even to humans (10, 11) and has been attributed to reduced risks in 2 major contributors to mortality, cancer and diabetes (12). It follows that better understanding of the molecular mechanisms underlying GH action has important implications to human health and disease.

Here we begin with a description of the physiological effects of GH on growth and metabolism, including the actions of the hormone on various tissues. Because alterations of gene expression profiles underlie hormone action, this review specifically focuses on molecular mechanisms of GH-mediated gene regulation, beginning with its signal transduction pathways and culminating with chromatin events at target genes. Given the combination of the critical role of IGF-1 in mediating the actions of GH and the body of experimental data investigating its regulation, we emphasize the Igf1 gene as a key target of GH activation.

Physiologic Actions of GH

GH is a 191-amino acid protein hormone secreted by the anterior pituitary gland that stimulates diverse anabolic actions throughout the body. Although multiple factors can influence its secretion, the primary hormonal control is via positive regulation by GHRH and negative regulation by somatostatin. In addition, sexual dimorphism in GH-secretory patterns has been well described, whereby males have pulsatile secretion but females exhibit a continuous secretion (13). As its name asserts, its most characterized role is in stimulating longitudinal bone growth. In addition, GH also impacts cell growth, differentiation, and metabolism, through coordinated actions on different tissues, including liver, adipose tissue, skeletal muscle, and bone. We begin by providing a brief review of the physiological actions of GH on these tissues, while highlighting known gene targets that underlie these effects (Figure 1).

Figure 1.

GH regulates gene expression in multiple tissues to carry out its physiological effects. Pituitary-derived GH travels through the circulation and influences target cells and tissues that express its receptor. Selected genes regulated by GH in liver, muscle, adipose tissue, growth-plate chondrocytes, and osteoblasts are highlighted here. Igf1 is a major transcriptional target of GH in multiple tissues. Liver-derived IGF-1 functions as an endocrine hormone, but local tissue-derived IGF-1 also has important physiological roles.

Liver

Liver is a major target organ for the action of GH, in large part because it is the major source of circulating IGF-1 (14, 15). Distinguishing actions of GH on growth and metabolism that are IGF dependent vs IGF independent is difficult, because the 2 hormones are intertwined in normal physiology (2, 16). Growth mediated by GH and IGF-1 is discussed in detail in a following section, but it bears mentioning here that the liver also contributes to the other carrier proteins that stabilize IGF-1 in the circulation. In particular, acid labile subunit is also stimulated by GH at the transcriptional level (17), and knockout of the Igfals gene further reduces serum IGF-1 and long bone length from the liver-specific IGF-I deficient mouse model (15).

Aside from growth, GH has significant effects on intermediary metabolism in liver (18). GH acts as a counterregulatory hormone to increase glucose levels in the setting of hypoglycemia by increasing hepatic glucose production via both glycogenolysis and gluconeogenesis. It is unclear which of these two exerts the predominant impact on hepatic glucose production, but a recent study has demonstrated a dose-dependent effect of GH on expression of Pepck, which encodes the key regulatory enzyme of gluconeogenesis (19). GH also effects lipid metabolism in liver through multiple mechanisms. A systems biology approach, combining metabonomics and pathway analysis of microarray expression profiling from liver in mice with alterations of GH signaling, revealed changes of lipid and choline metabolism with increased fat deposition (20). Three independent groups have generated mice with liver-specific knockouts for critical GH signaling proteins, GH receptor (GHR), Janus kinase (JAK)2, and signal transducer and activator of transcription (Stat)5, and observed a common phenotype of hepatic steatosis (21–23), demonstrating a key physiological role of GH in hepatic triglyceride secretion. The genes Cd36, Pparg, and Pgc1a have been implicated as among the primary transcriptional targets responsible for the steatosis (22, 24). GH also been shown to induce phosphorylation of sterol regulatory element-binding protein-1a, a transcription factor that directs lipid and cholesterol synthesis (25), and there is evidence to support that it also functions to increase hepatic fatty acid oxidation (26).

Other targets of GH in liver are genes of the cytochrome P450 (CYP) family, such as Cyp2c11, that function in hepatic steroid and drug metabolism (27, 28). The sexual dimorphism in GH-secretory patterns directs a sexually-dimorphic gene expression pattern that is characteristically noted in the CYP genes. In addition, expression profiling of human liver has also suggested that it may contribute to differences in lipid profiles that underlie sex differences in risk for cardiovascular disease (29). Finally, GH signaling is necessary for normal liver regeneration, with multiple mouse models of impaired GH signaling demonstrating reduced proliferation of hepatocytes following partial hepatectomy (23, 30).

Adipose tissue

The role of GH on body composition is apparent in the increased fat mass, preferentially visceral fat, in individuals with adult GH deficiency, which is responsive to GH treatment (31, 32). A primary role of GH on adipose tissue is in stimulating lipolysis, with an acute rise in free fatty acids regarded as a standard for assessing an appropriate in vivo GH effect (33). This effect is mediated, in part, by increasing hormone-sensitive lipase activity (34), but the specific mechanisms downstream of GH to achieve this have yet to be fully elucidated. Other proposed contributing mechanisms include alterations in the expression of the lipid droplet-associating protein, cell-death-inducing DFF45-like effector-A, and changes in secretion of adipokines, such as adiponectin (35–37). Aside from lipolysis, GH and IGF-1 have characterized roles in preadipocyte proliferation, differentiation, and senescence (37). In vitro differentiation of murine 3T3 preadipocytes to mature adipocytes is among the best characterized models of GH action (38, 39), with Fos and Jun noteworthy as 2 prominent target genes induced by GH in this process (40).

Skeletal muscle

GH and IGF-1 together promote skeletal muscle growth, maintenance, repair, and regeneration. Adults with untreated GH deficiency have decreased lean body mass, with improvements in muscle mass observed upon GH replacement (31, 41). Mouse models with impairments of the GH-IGF-1 axis characteristically have reduced muscle mass and function (3, 4, 42, 43), and there is strong evidence supporting contributions from GH-stimulated IGF-1 of both liver and muscle itself. In a background of pituitary GH deficiency, increase in lean body mass in response to exogenous GH in mice was attenuated in the setting of liver-specific knockout of JAK2 where serum IGF-1 is markedly reduced (44). Meanwhile, local Igf1 expression is increased in response to both injury and increased workload (45, 46), and forced expression of Igf1 in muscle can attenuate muscle loss of normal aging or pathologic loss in a model of muscular dystrophy (47, 48). The anabolic properties of these 2 hormones on muscle are characterized by increased protein synthesis and decreased protein breakdown. Interestingly, the metabolic effect of insulin resistance in muscle has been attributed directly to GH and not IGF-1; mice with muscle-specific knockout of GHR, but not IGF-1 receptor, demonstrate reduced 2-deoxyglucose uptake, reduced levels of insulin receptor, and increased serine phosphorylation of insulin receptor substrate 1, a modification that interferes with normal insulin downstream signaling (43).

Bone

Aside from the actions of the GH-IGF-1 axis at the growth plate to promote longitudinal growth, the axis also functions to regulate skeletal development and mineral acquisition (reviewed in Reference 49). A consistent finding in mouse models with disruptions of GH-IGF-1 axis is deterioration in parameters of bone health (14, 15, 44). A model with targeted disruption of GH signaling in liver discussed previously revealed that hepatic IGF-1 has a nonredundant role on GH-mediated acquisition of bone mineral density (44). No IGF-independent GH effect on bone has been defined (50), indicating that Igf1 is also a key effector of GH locally in bone. Conversely, local IGF-1 is regulated by other GH-independent mechanisms, including PTH (51). Experimental mouse models reveal that osteoblast-derived IGF-1 is a key determinant of bone mineralization. Targeted osteoblast-specific overexpression of Igf1 via the osteocalcin promoter produced a phenotype of increased bone mineral density and trabecular bone volume (52), whereas knockout of the gene in bone (and muscle) but not liver via Cre-recombinase expressed by the collagen type 1α2 promoter included a phenotype of reduced bone accretion (53).

GH Signal Transduction Pathways

As with other peptide hormones, GH acts upon a cell by binding its cognate receptor at the cell membrane. GHR is a member of the class I cytokine receptor superfamily with a single transmembrane domain. It exists as a preformed homodimer in its inactive state and subsequently undergoes conformational change upon binding of a single GH molecule to the high-affinity sites of each of the 2 GHR molecules (54). GHR itself has no intrinsic tyrosine kinase activity; however, the tyrosine kinase JAK2 is associated with the Box1 sequence in the cytoplasmic domain of the receptor and transphosphorylates following GH binding (55). Activated JAK2 then phosphorylates other tyrosine residues of the cytoplasmic domain of GHR. These phospho-tyrosine residues function as docking sites for intracellular proteins with Src homology 2 (SH2) or phospho-tyrosine binding domains and these proteins become targets for phosphorylation by JAK2 (56). Major signaling pathways downstream of GH are depicted in Figure 2.

Figure 2.

Multiple intracellular signal transduction pathways are activated by GH. GH (light blue) binds to the transmembrane GHR (dark blue) to initiate signaling pathways. Activated JAK2 (red) phosphorylates signaling molecules, including members of the Stat family (rectangular). Stats 1, 3, and 5a can participate in GH signaling (data not shown); however, Stat5b (dark green) is widely regarded as the principal transcriptional effector. JAK2 phosphorylation of Shc (green) begins a phosphorylation cascade of the MAPK pathway with activation of ERK (purple), whereas phosphorylation of IRS-1 triggers the PI3K (red) pathway. Noncanonical signaling independent of JAK2 through SRC (yellow) leading to activation of ERK has also been demonstrated. These pathways can then act on signaling effectors such as transcription factors (eg, C/EBPβ) to directly affect gene regulation or can lead to cross talk to impact the Stat pathway. Other inputs (shadowed arrow) may similarly alter GH signaling via multiple mechanisms including negative regulation of these pathways. IRS, insulin receptor substrate; c-Src, is the non-receptor protein tyrosine kinase.

Stat transcription factors

The best characterized targets of JAK phosphorylation are members of the Stat family of transcription factors (57). All Stat family members share a similar organization that includes a SH2 domain and a conserved tyrosine residue near the C terminus. Upon recruitment to the GHR-JAK2 complex, the Stat proteins are phosphorylated at the conserved tyrosine residue, dissociate from the membrane-bound complex, and form dimers with other phosphorylated Stat proteins via reciprocal interactions of SH2 domains and phospho-tyrosine residues. They then translocate to the nucleus where they bind to specific recognition sequences and influence gene transcription. Stats 1, 3, 5a, and 5b are all phosphorylated with GH signaling, and Stat5b is regarded as the major mediator of GH action. Aside from tyrosine phosphorylation in the canonical activation pathway, additional posttranslational modifications of Stat proteins include serine phosphorylation, acetylation, methylation, and sumoylation (56).

In addition to Igf1 (discussed in detail in a following section), Stat5b signaling is also directly implicated in the regulation of several other GH-stimulated genes including Igfals, Socs2, and Cish (58). Chromatin immunoprecipitation (ChIP) studies of rat liver have revealed Stat5b binding at consensus binding sequences in the promoters of these 3 genes with GH induction (59). Moreover, sexually-dimorphic liver gene expression profiles secondary to gender differences in pituitary GH secretion pattern also act principally through the transcription factor. Stat5b activity profiles recapitulate the sex-specific GH-secretory pattern (60), and targeted genetic loss of Stat5b essentially eliminates sexual dimorphism in hepatic gene expression profiles (61, 62). Yet it remains challenging to generalize about the extent to which a given pathway is responsible for downstream hormone signaling. Insights from comparing gene expression profiles in different experimental settings have challenged the concept that Stat5b mediates most GH effects, at least with regard to the number of genes altered at the transcriptional level. Rowland et al (63) generated a line of knock-in models of GHR with targeted mutations or truncations of the cytoplasmic tail and observed that a central segment of the cytoplasmic domain of GHR (amino acid residues 391–569) was required for Stat5 activation Expression profiling by microarray analysis revealed that more transcripts were altered by deletion of GHR or a shorter truncation that abolished JAK2 activation than with the truncation producing loss of activation of Stat5. Likewise, Vidal et al (64) also concluded that most transcripts that are increased with GH signaling are not dependent upon Stat5b, based upon their analysis that revealed only 20% of GH-regulated transcripts declined in abundance in the setting of a dominant-negative Stat5b. Admittedly, the study did not assess residual Stat5 activity in the experimental paradigm, and it is possible that expression of some true Stat5b-dependent genes was maintained with some residual Stat5 activity. Whereas the prototypical role of Stat5b action is to activate transcription, there are also multiple reports of its role in transcriptional repression (63–65). In contrast to the description that most GH-stimulated genes are not Stat5b dependent, most genes acutely repressed by GH are Stat5b dependent. Analysis of the above microarray data set with a dominant-negative Stat5b for transcripts that were acutely reduced with GH determined that more than 90% were Stat5b dependent (66).

Other signaling pathways

Two well-known signal transduction pathways that also act downstream of GH are the MAPK and the phosphatidylinositol 3′-kinase (PI3K) pathways. Conventional models hold that these pathways are activated by tyrosine phosphorylation of a signaling molecule by JAK2. With the MAPK pathway, the adaptor protein Shc is phosphorylated by JAK2 and then acts via Grb2 and SOS to signal a phosphorylation cascade through Ras, Raf, MAPK kinase, and ERK (67). The PI3K pathway is activated through phosphorylation of insulin receptor substrate 1 (68), thus using the same effectors as insulin- and IGF-1 in signal transduction. In comparison with the Stat proteins, there has been less elucidation of details of the transcription factors that are activated by GH via the MAPK and PI3K pathways and then bind to regulatory domains to influence gene expression. A notable exception to the above is the transcription factor CCAAT/enhancer-binding protein (C/EBP)β that plays a central role in GH-induced adipocyte differentiation (69). A model gene for characterizing mechanisms of GH-regulated gene transcription by C/EBPβ has been Fos, a protooncogene that is rapidly and transiently stimulated by GH, the gene product of which can function as a transcriptional effector of GH (40, 70). Among the characterized posttranslational modifications of C/EBPβ are phosphorylation of a threonine residue in the regulatory domain by MAPK and acetylation of a lysine residue in the transcriptional activation domain by p300 (71, 72). It is unclear whether GH directly impacts p300 activity in this system, but C/EBPβ and p300 are found to occupy the Fos promoter as a single complex upon GH treatment. In addition, genome-wide analyses have identified associations of local C/EBPβ and Stat5 binding in 3T3-L1 cells (73). Altogether, the comprehensive studies of activation of the Fos promoter have provided valuable insight about how multiple mechanisms of GH action can integrate to influence gene expression. Finally, an additional adaptor protein that is phosphorylated by JAK2 with GH is SH2B1β (74). The activated protein functions in actin cytoskeleton remodeling that underlies GH-dependent macrophage mobility (75), illustrating another of the diverse effects of GH.

Noncanonical GH signaling

Although canonical signaling downstream of GHR requires activation of JAK2, there has been strong experimental evidence of other JAK2-independent signaling pathways. Activation of the Src family kinases by GH leads to a signaling cascade that includes activation of ERK. In a mouse model with targeted mutations of the Box 1 sequence of GHR to which JAK2 associates, investigators observed absence of activation of hepatic JAK2, Stat3, Stat5, and Akt in response to systemic GH, but preserved activation of Src and ERK (76). Although the phenotype of this mouse model appears indistinguishable from that of Ghr−/− mice, gene expression profiling revealed differences among the two that are hypothesized to reflect preserved downstream signaling through ERK. Conversely, a different mutation of GHR prevents the conformational change necessary for ERK activation but maintains JAK2 activation (77). These data support a model in which distinct conformational states of GHR alter the ratio of activation of JAK2/Stat and Src/ERK in response to GH; however, data from other systems have indicated that JAK2 is necessary for full GH-induced ERK activation (78), and therefore the relevant signaling pathway may be cell- and tissue specific.

There have also been several studies investigating the potential role of nuclear localization of GHR in mediating physiological effects. Nuclear translocation of membrane receptors, eg, epidermal growth factor receptor and fibroblast growth factor (FGF) receptor, is a well-described feature that has been associated with cell transformation (79). Targeting GHR to the nucleus is sufficient to induce tumorigenesis and tumor progression in in vivo models (80), and the extracellular domain of the receptor corresponding to GH-binding protein has been shown to have transcriptional activity in reporter assays (81). Increased nuclear GHR accumulation has been reported in several cancers (82), as well as in rapidly proliferating hepatocytes in a mouse model of liver regeneration (80), suggesting that this is not strictly an experimental phenomenon; however the contribution of nuclear GHR to gene regulation in normal physiology remains to be established.

Negative regulation of GH signaling

Much has been written about the multiple mechanisms that function to negatively regulate GH-signaling pathways (reviewed in Reference 83). GH binding enhances the rate of internalization of GHR through clathrin-coated pits and cavelolae in a process that is ubiquitin dependent (84). The internalized GHR is then degraded via either proteasomes or lysosomes (85), thereby limiting GHR abundance and decreasing GH responsiveness of the cell. Protein tyrosine phosphatases (PTPs) remove the activation mark placed by JAK2 on its targets, and several PTPs have been implicated in down-regulation of GH signaling including small heterodimer partner-1 and -2, PTP1b, and PTP-H1 (86–88). Direct interactions of JAK2 and PTP1b have been observed via phospholipase Cγ1, providing a mechanism for finely balanced control of GH action (89). Members of the suppressors of cytokine signaling (SOCS) family play a major role in the negative regulation of JAK/Stat signaling, including downstream of GH. All SOCS proteins have an SH2 domain that allows interactions with phosphotyrosine residues, but they employ diverse mechanisms to disrupt signaling including competitively masking binding sites of the GHR-JAK2 complex that recruit Stats and adapter proteins, inhibition of JAK2 kinase activity, and ubiquitin-ligase action that contributes to proteasomal degradation of GHR and JAK2 (83). Knockout of the gene for SOCS2 in mice produces a phenotype of overgrowth, and therefore it is regarded as a key negative regulator of growth (90). As noted above, GH stimulates transcription of several members of the SOCS family via Stat5b, demonstrating an integrated feedback mechanism (58–60, 64).

At the level of chromatin, the transcriptional repressors BCL6 and CUX2 have been characterized to repress a subset of GH-activated and Stat5b-dependent genes, largely in the context of sexually dimorphic gene expression patterns in liver (91, 92). BCL6 is preferentially expressed in males, whereas CUX2 is highly female specific. In contrast to genes of the SOCS family that are activated with GH, the Bcl6 gene has been shown to be repressed by acute GH in an in vitro model system (93). BCL6 binds to a consensus sequence with significant overlap with that of Stat5b (94, 95), and a transition in occupancy from BCL6 to Stat5b at regulatory domains of many GH-induced genes, including Cish and Socs2, has been demonstrated (59, 91, 96). In genome-wide studies, more than 50% of BCL6-binding sites overlapped with Stat5-binding sites, with a significant enrichment at female-biased genes, suggesting that it plays a role in active repression of female-biased genes in males (91). In comparison, CUX2 recognizes an independent consensus motif, yet its binding sites are also enriched near Stat5-binding sites, and directs transcriptional repression of male-biased genes but transcriptional activation of female-biased genes (92).

Cross talk with other signaling pathways

Identification of the key downstream signaling molecules of GH signaling naturally provides mechanisms by which cross talk with other signaling pathways can be achieved. Selected examples of how other hormones or environmental states influence gene regulation by GH are highlighted here. Oral, but not transdermal, estrogen attenuates the action of GH in liver, including in stimulating IGF-1 production (97), with a key mechanism the stimulation of Socs2 expression in liver by estrogen acting via estrogen receptor-α (98, 99). Direct cross talk of GH and IGF-1 downstream signaling has also been investigated. A complex of GHR-JAK2-IGF-1R has been described in several cell types (100–102), and further studies have indicated that IGF-1R perpetuates GH signaling by preventing PTP1b-mediated suppression of Stat5 activation (103). Starvation and malnutrition lead to a state of GH resistance that can be considered an adaptation of fasting (104). The hormone FGF21 is induced by fasting, and a characteristic feature of a transgenic mouse model with increased levels of FGF21 is growth inhibition (105). Molecular studies of this model reveal reduced hepatic Stat5 phosphorylation and Igf1 expression, with a concomitant increase in Socs2 expression. The mechanisms by which FGF21 impacts Stat5 phosphorylation are still under investigation, but signaling via its coreceptor βKlotho has been shown to be necessary in genetic mouse models (106), and increased expression of leptin receptor overlapping transcript and leptin receptor overlapping transcript-like 1, proteins that regulate intracellular protein trafficking, has also been implicated as a contributing factor (107, 108). In addition to its actions at the liver, FGF21 can also act at the level of chondrocytes of the growth plate to achieve its growth-inhibitory effect. FGF21 blocks GH-stimulated chondrocyte thymidine incorporation and collagen X expression in a dose-dependent manner (108, 109). The histone deacetylase SIRT1 is also activated with fasting and has been shown to directly interact with Stat5 (110). Several lysine residues that lie near the conserved tyrosine phosphorylation site of Stat5 can be acetylated, and, upon interaction, SIRT1 deacetylates Stat5 and diminishes its transcriptional activity. Because growth inhibition is a characteristic consequence of glucocorticoid excess, it is unexpected that hepatic-specific deletion of the gene encoding the glucocorticoid receptor results in an approximately 30% decrease in growth (111). Glucocorticoid receptor can directly interact with an N-terminal domain of Stat5 and has additionally been detected at Stat5-binding sites of GH-regulated genes, suggesting that it has a role as a coactivator of Stat5 under physiological levels of glucocorticoids (112). Conversely, orphan nuclear receptor small heterodimer partner is characterized as a corepressor of Stat5, because it likewise interacts with the transcription factor, but instead leads to diminished GH-stimulated, Stat-mediated gene expression impacting metabolism, such as at the Pepck gene (19). Finally a meta-analysis of genome-wide Stat5 binding in liver identified significant overlap for local binding of hepatocyte nuclear factor (HNF)-4A, C/EBPα, FoxA1, and FoxA2, although direct interactions or other mechanisms for cross talk were not assessed in the study (73). An emerging concept in gene regulation is clustering of multiple transcription factor binding sites at cis-regulatory modules (113, 114), and the association of Stat5 binding with other transcription factor-binding sites in specific cell-types begins to provide additional clues to mechanisms of GH-regulated gene transcription.

Regulation of IGF-1 Transcription by GH

Since the formulation of the somatomedin hypothesis by Salmon and Daughaday (1) more than 50 years ago, Igf1 has been established as a primary physiological target of GH action, and therefore the mechanisms underlying GH-induced Igf1 gene transcription are a topic of fundamental interest. Whereas the Igf1 gene organization and mechanisms of regulation are complex, the IGF-1 protein is a straightforward single-chain 70-amino acid peptide that bears structural similarity to insulin (Reviewed in Reference 115). In rodents and humans, the gene consists of 2 promoters and 6 exons and extends over 70–80 kb of genomic DNA. Interestingly, more than 100 total distinct mRNAs have been described, with specificity of detection with regards to tissue and developmental stage. These multiple primary transcripts arise from multiple transcriptional initiation sites of both promoters, alternative splicing, and differential polyadenylation, but all are translated to only 2 precursors that are processed to the identical peptide hormone. P1, the major promoter, is active in all tissues, whereas P2 activity is primarily restricted to liver. Transcripts originating from P2 contribute to only 25% of the transcripts in liver (116, 117), yet this fraction has been suggested to produce phenotypically significant outcomes, as exemplified by variations of serum IGF-1 levels attributed to differences in P2 activity between 2 inbred mouse strains (118).

Initial characterization of key regulatory features of the Igf1 gene focused on its 2 promoters, and mapping key regions necessary for basal promoter activity. Roles for several liver-enriched transcription factors were thus characterized, including promoter binding sites for C/EBPα, HNF-1α, and HNF-3β/FoxA2 (119–121), yet they did not provide additional insight about the mechanisms for GH activation. For example, although Thomas et al (122) identified 6 distinct DNA-protein interaction sites of proximal P1 and exon 1 of the rat Igf1 gene, GH treatment did not alter any of these interactions, indicating that they may participate in basal promoter activity but did not dynamically increase transcription upon GH treatment. In contrast, a single GH-responsive deoxyribonuclease (DNase)-I hypersensitivity site termed HS7 had been identified in the second intron of the gene, 3.7 kb from the exon 1 transcriptional start site (123, 124).

In the past 10 years, several lines of evidence together established the transcription factor Stat5b as the key intermediate linking GH to Igf1 transcription. It had previously been well recognized that Stat5b was phosphorylated by JAK2 with GH activation; however, it was not otherwise distinguished from other factors downstream of GH-signaling pathways. To directly address the potential role of Stat5b in GH-induced Igf1 transcription, Woefle et al (125) generated recombinant adenoviruses expressing modified variants of Stat5b with either dominant-negative or constitutively-active properties. Using the hypophysectomized rat model, they observed that GH-induced Igf1 gene transcription in liver was blocked by the dominant-negative Stat5b, whereas the gene was expressed even in the absence of GH with the constitutively-active form, indicating that activated Stat5b was both necessary and sufficient to drive Igf1 transcription with GH. The concurrent identification of a homozygous mutation in the gene encoding Stat5b in a female patient who presented with profound short stature, GH insensitivity, and an immune deficiency provided strong genetic evidence for its critical role (126). Since then, more than 10 additional patients have been described with an essentially identical clinical phenotype (127).

Characterizing Stat5b-binding enhancers of Igf1

Establishing Stat5b as the critical intermediate in regulation of Igf1 expression by GH set the stage for elucidating chromatin events that occur with transcription, with initial efforts focused on identifying key Stat5-binding enhancers. Woelfle et al (128) evaluated the intronic HS7 domain mentioned above, with its property as a DNase I-hypersensitivity site induced by GH and containing tandem consensus Stat5-binding motifs, as a potential candidate. Using a combination of in vivo ChIP, gel-shift assays, and cell-based reconstitution experiments, they demonstrated that Stat5b bound to the HS7 domain, and that the domain conferred GH responsiveness to the major Igf1 promoter, thereby implicating it as the first characterized Stat5-acting cis-regulatory element of the gene. Three independent groups extended the list of Stat5-binding domains of the Igf1 gene, typically narrowing candidates to conserved regions containing consensus Stat5 binding sequences as a starting point (60, 129, 131). Altogether, at least 7 distinct Stat5-binding domains scattered across a 200-kb segment encompassing the gene in the mouse and rat genomes were identified in these studies, with a significant proportion containing 2 or more consensus-binding sequences. Interestingly, no Stat5-binding domains were found proximal to either of the promoters, consistent with other experiments that have found only modest changes in proximal promoter activity with GH (128).

Although these studies have provided some insight to regulatory domains of the Igf1 gene, there remain key limitations. First, the studies may have limited sensitivity in that they relied on the assumption that biologically relevant binding sites would have an evolutionarily conserved consensus Stat5-binding sequence. Recent reports investigating genome-wide transcription factor binding across species have revealed remarkable divergence, suggesting that the above assumption is largely flawed (132). Indeed studies from our laboratory using human liver have suggested that homologous Stat5-binding consensus motifs do not display open chromatin as had been observed in mice (133). Nevertheless, results of an unbiased Stat5 ChIP-seq study in mouse liver confirm that the most robust Stat5-binding sites of the Igf1 locus had been appropriately captured by the candidate domain approach (91). Next, defining the genomic location of putative enhancers for these studies was largely arbitrary, with candidate domains restricted to approximately 200 kb of the Igf1 gene locus. A recent unbiased search for associations of transcription start sites and distal elements by chromosome capture conformation carbon copy (5C) in 3 different cell lines demonstrated that only 7% of looping interactions occur with the nearest gene (134). Moreover, long-range interactions between enhancers and promoters occurred most frequently from a distance of 120 kb upstream from the transcription start site, a location that had not been directly assessed in any of these 3 studies. None of the characterized Stat5 domains has been validated to interact with either of the Igf1 promoters experimentally, and therefore the concept that Stat5b binding at these domains leads to enhanced transcription from the promoters is an inference based upon the genomic location of the domains, chronology of detecting acute changes, and modeling in cell-based reporter assays. A prominent example in the literature of how these types of inferences may be misleading is with the single-nucleotide polymorphism (SNP) most strongly associated with obesity that lies in an intron of the FTO gene (135). Although there was a strong rationale that the SNP altered FTO expression to impact the phenotype, recent experiments indicate that it achieves its effect by instead influencing expression at the IRX3 gene 500 kb away (136). Occupancy of transcriptional proteins p300, Med1, and RNA polymerase II at a subset of Stat5-binding domains and acute changes in local chromatin marks including H3 and H4 acetylation and mono- and trimethylation of H3K4 with GH have been reported (131), but there remains limited data to propose models of the chromatin events that occur with GH-induced transcriptional initiation. Note that the unbiased Stat5 ChIP-seq study demonstrates genome-wide peaks but provides no additional insight about which Stat5-binding sites are interacting with the Igf1 promoters to enhance transcription. Finally, it seems highly likely that multiple enhancers participate in transcription, and whether they behave hierarchically, synergistically, or redundantly remains unknown.

The potential clinical significance of defining gene enhancers is illustrated in 2 recent reports focused on pancreatic islet cell development. Weedon et al (137) demonstrated that recessive mutations in a developmental enhancer 25 kb downstream of a gene encoding a pancreas-specific transcription factor are the most common etiology of isolated pancreatic agenesis. Meanwhile, a large team of investigators has characterized the human islet “cistrome,” encompassing genome-wide chromatin features and transcription factor-binding sites in this important cell type (138). Among the key findings of the study are that that sequence variants associated with type 2 diabetes often map to the enhancers corresponding to a cluster of transcription factor-binding sites. This follows from a previous study that demonstrated that other identified disease-associated SNPs frequently map to gene-regulatory domains that are characteristically tissue specific (139). By analogy, one could envision that both genetic and epigenetic changes at enhancers of the Igf1 gene may underlie clinical scenarios of impaired hepatic IGF-1 production, such as commonly observed in individuals of short stature (140).

Tissue-specific mechanisms of Igf1 gene regulation

To date, liver has been the primary tissue in which GH-regulated Igf1 gene transcription has been studied most extensively. Igf1 expression in liver markedly exceeds that in any other tissue, and most serum IGF-1 is known to originate from liver (14, 15, 141). Our group has recently explored tissue specificity of the characterized regulatory elements of the Igf1 gene by using the technique of formaldehyde-assisted isolation of regulatory elements to enrich for domains of open chromatin in different primary mouse tissues (133), and our directed analysis validated results from a genome-wide study of DNase hypersensitivity sites in mouse liver (142). We demonstrated that promoter 1 is open in liver, kidney, and spleen, but that open chromatin at the Stat5-binding domains is restricted to liver. Furthermore, chromatin accessibility at either promoter 1 or the Stat5-binding domains in liver was not dependent upon GH signaling, because we identified open chromatin at these regulatory domains even in the Ghrhr−/− mouse with congenital absence of GH. These findings have led us to speculate about how key regulatory domains of the Igf1 gene are marked during development. We hypothesize that a hepatic-specific developmental program involving a set of transcription factors sets the stage for specific domains to later function as GH-responsive cis-regulatory elements and, further, that disruption of this program could present with GH resistance. Pioneer transcription factors in other settings, including at hepatic-specific genes such as Alb encoding albumin, are well described (143). Likewise, the genome-wide enrichment in HNF4A, C/EPBα, FoxA1, and FoxA2 binding with Stat5 binding in liver has previously been discussed (73), but no specific overlap at regulatory domains of the Igf1 gene has been highlighted. Meanwhile, Igf1 is expressed in nearly all tissues, and GH stimulation of Igf1 gene expression has also been described in other tissues. For example, GH has been shown to acutely increase Stat5 phosphorylation and Igf1 expression in muscle both in vitro and in vivo, including in human muscle biopsies (144–146). Similar findings have been reported in adipose tissue in rat (147). Yet it remains well acknowledged that few cell culture systems maintain GH-stimulated Igf1 expression, thereby providing few experimental models. Working with other primary tissues can be technically challenging, likely contributing to the relative scarcity of studies of hormone-regulated transcription derived from other tissues.

An important physiological tissue for understanding of GH-regulated Igf1 gene expression is the growth plate. Whereas the original somatomedin hypothesis had proposed that an endocrine factor later identified as IGF-1 was responsible for mediating all of the anabolic properties of GH (1), the demonstration of accelerated bone growth upon direct administration of GH to the growth plate supported the counter explanation that locally derived IGF-1 elicited the growth effect (148). Defining the physiological roles of liver-derived endocrine-acting IGF-1 vs local tissue-derived paracrine- and autocrine-acting IGF-1 is challenging, because normal feedback mechanisms largely prevent manipulation of one hormone in the axis without impacting the other (2, 14). The current consensus based upon rodent models is that both endocrine and local IGF-1 impact growth (149, 150), with the group of Stratikopoulos et al (149) estimating the total contribution of endocrine IGF-1 as 30%, local IGF-1 as 39%, with the remainder independent of IGF-1. There have been few reports exploring the mechanisms of Igf1 regulation in chondrocytes of the growth plate or osteoblasts in bone. Older studies investigating Igf1 expression in growth plate chondrocytes and the effect of GH have primarily been restricted to in situ hybridization for Igf1 transcripts and immunohistochemistry for IGF-1 with in vivo models such as the hypophysectomized rat (151–153). More recently, Stat5 has been shown to be acutely phosphorylated in response to GH in growth-plate chondrocytes (154), consistent with the presumption that Stat5 also participates in Igf1 transcription at the growth plate based upon the similar growth phenotypes in individuals with mutations of the GH receptor and Stat5b genes (127). Moreover, a role for nuclear factor (NF)-κB p65 has been also revealed in a series of experiments by Wu et al (155). Their group demonstrated that both inhibition of nuclear factor-κB (NF-κB) activity by the inhibitor pyrrolidine dithiocarbamate and depletion of NF-κB by targeted small interfering RNA eliminated the effects of GH on Igf1 expression and chondrocyte proliferation and differentiation. Additional studies further suggest that the impact of NF-κB on GH signaling is not limited to chondrocytes, because a patient with a heterozygous mutation of IκBα demonstrated clinical features of GH resistance including poor growth with low levels of liver-derived serum IGF-1, and skin fibroblasts derived from the patient also maintain this phenotype (156).

Most literature on osteoblast expression of Igf1 has focused on the physiological role of locally produced IGF-1 on bone mass acquisition (49), whereas there have been limited studies regarding the mechanisms of regulation. Reductions in trabecular bone remodeling were observed in a mouse model with loss of GHR, but this finding was not observed with loss of Stat5, suggesting that GH does not act via Stat5b in bone (157). Cross talk of downstream GH and IGF-1 signaling that has been discussed previously had been first characterized in osteoblasts (101), but it does not appear that this is unique to this tissue. Smith et al (158) recently reported on 2 micro-RNAs that are up-regulated during osteoblastic differentiation that bind to the 3′-untranslated regions of Igf1 transcripts and function in negative regulation, adding to previous reports of other micro-RNAs targeting Igf1 transcripts in other contexts (159). Similarly, the deadenylase protein Nocturnin interacts with the 3′-untranslated region of Igf1 mRNA to suppress gene expression in bone (160). Finally C/EPB-δ has been shown to regulate Igf1 transcription in osteoblasts by binding to an element in the proximal promoter (161). Activation of this transcription factor occurs downstream of PTH and cAMP (130, 162), and no relationship with GH signaling has been described.

Summary and Future Directions

The broad physiological actions of GH and IGF-1 on growth and metabolism and extending to aging support the detailed study of mechanisms of GH gene regulation. Much has been learned about the signal transduction pathways that are activated by GH, yet there remain key questions to address in this field and experiments assessing events in chromatin are under represented. Tissue specificity in mechanisms of gene regulation has been highlighted in this review and can be manifest at many different levels. For example, other signaling molecules that cross talk with GH signaling pathways may be restricted to specific tissues, or differences in the composition of transcriptional cofactors may influence gene expression profiles. The recent literature has accentuated the role of gene enhancers in contributing to disease states, and tissue-specific chromatin context can dictate what regulatory domains in genes are active. Understanding the mechanisms by which enhancers are established during tissue differentiation may provide an important platform for their study. Although Igf1 is the most critical effector of GH and, hence, the most studied, characterization of its key regulatory domains remains incomplete, with virtually no current insight about the domains that function in humans or those active in tissues other than liver. Finally, strategies to modulate signaling pathways to target specific genes would be desired to allow enhancement of selected GH effects while not inducing any harm.